My workaround is log into the machines to run openclaw gateway restart. I guess I can setup a systemd timer, but is this behaviour expected? Is there a bug tracking this? It's happening on two independent uptodate bots for me.
#WhatsApp stops working
rotund cradle